Print KL/CL/KE events for all warps (#1544)
* Print KL/CL/KE events for all warps
* Fix count off-by-one issue
* Fix opCount in KE and restore CPU thread option
* Simplify count calculation
[ROCm/rccl commit: ebf7e2305e]
이 커밋은 다음에 포함됨:
@@ -419,7 +419,8 @@ struct ncclCollTrace {
|
||||
uint8_t type;
|
||||
uint8_t bid;
|
||||
int16_t funcIndex;
|
||||
uint32_t data_0;
|
||||
uint32_t data_0:24;
|
||||
uint8_t tid;
|
||||
uint8_t channelId;
|
||||
uint64_t timeStamp:56;
|
||||
union {
|
||||
|
||||
새 이슈에서 참조
사용자 차단